c - Optimizing an openCL calculation with bitwise operations -


let me commence stating objective: attempting compare 2 13-bit integers: >, <, , =; have calculation trillions of times, imperative optimize as possible.

my program utilizes python 2.7 , pyopencl achieve calculation. averaging 800 gflops out of ati radeon 6870 fine now.

so here question: if instead of caring out <, >, , = operators on 4 byte floats (as doing now), wrote bitwise functions handle <, >, , =, , able process 2 - 13 bits objects @ time, increase speed? or c have efficient way of finding <, , >, , = (obviously) floats?


Comments

Popular posts from this blog

Django REST Framework perform_create: You cannot call `.save()` after accessing `serializer.data` -

Why does Go error when trying to marshal this JSON? -